T.R | Title | User | Personal Name | Date | Lines |
---|
127.1 | MCC$SYTEM, MCC$COMMON & MCC$SPECIFIC | TOOK::DITMARS | Pete | Wed May 09 1990 15:13 | 57 |
| Hi,
******************************** WARNING!! ************************************
I can't make any statements about the advisability of redefining your MCC$
logicals. I'm pretty sure the current MCC vmsinstal procedure will not be
happy with you doing so, and you will be bitten, one way or another, during
upgrades unless you "un-do" whatever you've done before doing an upgrade.
This stuff is not in my realm of responsibility, but I do understand
enough about the use of search lists to (hopefully) help answer your question.
If this is something lots of folks will want to do, maybe we (MCC) should
explicitly add support for the kind of thing you are trying to do.
*******************************************************************************
Your problem isn't with MCC$COMMON, it's with MCC$SYSTEM. There's some good
information on clusters and common/specific stuff in the VAXcluster manual of
the VMS docset.
The MCC$COMMON logical is used to place (create) files that MUST be common,
but the MCC$SYSTEM logical is used to access most files. Your current
definition of MCC$SYSTEM:
"MCC$SYSTEM" = "SYS$SYSROOT:[MCC]"
relies on your definition of SYS$SYSROOT. Normally VMS systems define
SYS$SYSROOT to be a search list, the first element of which points to a
system-specific directory, the second element of which points to a
cluster-common directory. For example, on node FOO in a cluster where FOO is
system 11, you would have something like:
"SYS$SYSROOT" = "DISK$DUA0:[SYS11.]" (LNM$SYSTEM_TABLE)
= "SYS$COMMON:"
1 "SYS$COMMON" = "DISK$DUA0:[SYS11.SYSCOMMON.]" (LNM$SYSTEM_TABLE)
What that all means is that your stuff in your private MCC$COMMON was never
being seen, but the stuff in SYS$COMMON:[MCC] was.
What you want is something like this:
"MCC$COMMON" = "DUA1:[MCC_OLDDIR]"
"MCC$SPECIFIC" = "SYS$SPECIFIC:[MCC]"
"MCC$SYSTEM" = "MCC$SPECIFIC:"
= "MCC$COMMON:"
To build a logical name search list, separate each element by a comma in your
define command, like so:
$ define/system mcc$system mcc$specific:, mcc$common:
If you try this, *then* we'll see if there's any hard-coded stuff in your way.
NOTE: I would NOT consider re-defining SYS$SYSROOT
regards,
Pete
|
127.2 | untangle the mess & dictionary MUST be movable | GDJUNK::HOULE | Steve, NM is the future! | Wed May 09 1990 17:55 | 16 |
| Thanks for the help. I'll have to untangle the mess I have. It seems MCC is
using my redefined logical. For example, thats where all my alarms rules went.
I'm not a VMS expect but I have found that with Many applications if a logical
is defined in the startup file, its there for the very purpose I tryed to use
it for.
***I Strongly believe that Users MUST be at LEAST be able to put the dictionary
file on a disk other than the SYSTEM disk!!!!!!!!!!
No matter what platform MCC recommends a "large" application like MCC can't
require such Huge files on the system disk; Look at what's already happening,
a 150Mg disk is becoming to small to support VMS & DECwindows (a basic OS)!
Also, when the performance FM comes to life its files will need to be on
other disk.
Thats my two cent, ===Steve
|
127.3 | try search lists | TENERE::DUNON | Paul Dunon - Telecom Engineering - VBO | Thu May 10 1990 07:11 | 16 |
| Hi,
We have installed MCC on a cluster but it can only be started on some sattelite.
I've modified my satellite mcc$startup.com the following way:
$ define/system mcc$system disk$alezan_64750:[mcc],sys$sysroot:[mcc]
$ define/system mcc$common sys$common:[mcc]
$ define/system mcc$specific disk$alezan_64750:[mcc]
to have the dictionary, gs_pool, event_pool and ptb_parser data files on my
local disk when I run DAP, PTB and MCC.
It works fine.
-- Paul
|
127.4 | what are logicals for | GOSTE::CALLANDER | | Thu May 10 1990 15:28 | 32 |
|
We seem to be walking around, without discussin, "What does MCC
use logicals for, and which ones can I redefine?"
Okay let's clarify what is being asked here. I frist saw a question
about storing icons for the IFT release of the Iconic Map PM; then
there was a question regarding the location of the pool, dictionary,
and parse tables.
MCC$ICON is use as the logical for identifying the location of the
icons. You should be able to do a show logical on this and find
it (it may be MCC$ICONS, you can check with TOOK::F_Messinger, or
TOOK::D_Moore).
MCC$SYSTEM is a search list used to find all of the dictionary files,
the parse tables and the pools. Please note though that moving one
of the dictionary files to a new directory and resetting the logical
will NOT do it. You must move all of the MCC$... dictionary files to
"move" a dictionary. In the developers toolkit manuals there is
information on how to define the logicals and move files to create
private libraries.
Regarding your cluster problem, I run on a cluster now, the only
thing we have to do is make sure that the DCLtables are reinstalled
on each node so that the commands are available. If you don't do
this then it tries to run the previous version (which was probably
deleted by the installation procedure).
Does this help?
|
127.5 | It shouldn't be a big problem to solve | NSSG::R_SPENCE | | Mon May 14 1990 23:35 | 10 |
| We have been able to take the whole [MCC...] directory set to another
disk and simply update the startup file to make sure ALL the required
logicals pointed to the new disk. It seems to work and (in my humble
opinion) MUST work for the released V1.0.
Actually I had to hack the KITinstal to actually install it that way
but that's another storey. I did write up my comments and give them
and the modified code to the KITINSTAL developer.
s/rob
|